Effective business intelligence: From decision support to supporting decisions

BI systems know what you know if you know how to use the tools

Howard Dresner of GartnerGroup (www.gartner.com) is credited with coining the term business intelligence (BI) to describe applications that interpret historical data, spot trends and measure performance.

"BI improves the quality and time lines of decision making," he said. "When business users can obtain rapid answers to questions they could never ask before, BI becomes a strategic weapon."

BI systems separate online transaction processing from online analytical processing (OLAP) and support the access, management, analysis and presentation of structured data. Therefore, data must be structured to be useful for BI. Of the technologies involved in BI, OLAP, data warehousing and data mining are the most critical. When integrated, they form a complete decision analysis system.

While BI is recognized for processing structured data for decision-making support, those systems are heading toward processing both structured and unstructured data. "BI solutions have demonstrated the ability to help enterprises uncover, analyze and understand information derived from a wide range of data sources," said Robert Moran, VP of decision support research at Aberdeen Group (www.aberdeen.com).

Beyond OLAP, which spots trends such as "How many red trucks do we sell on Tuesdays?" is MOLAP (multidimensional OLAP), which offers multidimensional data analysis. That multidimensional viewing allows a company to generate reports simultaneously from various perspectives (e.g. customer, financial, product) giving comparative perspectives.

This "slice and dice" analysis is key for strategic planning, but can also be used by department managers in the course of their daily work. "People make decisions by comparing things," said Nigel Pendse, a principal with OLAP Solutions (www.olap.csc.no). He describes BI as "really about being able to visualize a business from many different angles."

"Most analysis is multidimensional by nature," agreed Karen Rubenstrunk of Meta Group (www.metagroup .com). "So are most business problems."

Relational OLAP (ROLAP) handles larger data sets of relational data, but places larger demands on performance. A cross between MOLAP and ROLAP is offered as hybrid OLAP (HOLAP), which balances performance demands with larger data sets.

Ultimately, the information derived from all that processing must be put to use. Historically, the processing itself was so complicated that by the time the reports were generated, they were outdated. Worse, they didn't offer the opportunity to refine the queries based on the information they provided.

Accessing those reports was expensive and cumbersome, reserved more for senior management who might receive a printed analysis. To create those reports, IT had to "lock the processes down" to generate a report. It was too taxing on system administrators to consider creating a new report to answer every "what if."

"We built walls around it," said Mark Moorman, global manager with SAS (www.sas.com), "but on the business side, information changes. Answers to questions invariably drive the next question."

To generate a new report or analyze data from a changed perspective, the analytical processing itself needed to be flexible. The solution has come gradually in the form of data warehouses. Information stored in repositories, which can include multiple databases, corporate intranets and daily ledgers, can now be accessed and analyzed on the fly.

BI allows progressive queries by following Dresner's and others' tenets:

• Explore data to identify trends or find uniqueness in information.
• Quantify to place value on relationships and verify the perceived trends.
• Qualify the results by placing them in context.

OLAP and multidimensional database tools are critical pieces of that process by largely satisfying the first two tenets. The third brings business intelligence into the realm of knowledge management by putting the analysis in the context of both related information and experience.

According to Moorman, that allows organizations to anticipate rather than react to change. "This avoids spike-based decision making based on the results of the latest report," he said, "whatever passed your desk last."

The ability to quickly analyze a business from new perspectives helps organizations react to change as it occurs. Faster reaction time often distinguishes a company from its competitors.

"Ten years ago a product or service gave you a competitive advantage because it took a while for competitors to adapt," said Mark Gaydos, a product manager with BackWeb (www.backweb.com). "Today companies adapt much faster, creating duplicate products and services more quickly. Once products and services become identical, they distinguish themselves through one word--information."